
# Format Python code structure
%load_ext nb_black

# Numpy & Pandas to import, process & understand data
import numpy as np
import pandas as pd

# Libraries to help with data visualization
import matplotlib.pyplot as plt
import seaborn as sns
import warnings

# setting graph plots background style as darkgrid
sns.set_theme(style="darkgrid")
# to generate graphs/plots inline in jupyter notebook
%matplotlib inline
# to round numerical columns to 3 digit decimal
pd.set_option("display.float_format", lambda x: "%.3f" % x)


# Removes the limit for the number of displayed columns
pd.set_option("display.max_columns", None)
# Sets the limit for the number of displayed rows
pd.set_option("display.max_rows", 200)

# to split the data into train and test
from sklearn.model_selection import train_test_split

# to build linear regression_model
from sklearn.linear_model import LinearRegression

# to check model performance
from sklearn.metrics import mean_absolute_error, mean_squared_error, r2_score


# I'm setting the random seed via np.random.seed so that
# I get the same random results every time
np.random.seed(1)

# Turning off the warning SettingWithCopyWarning
pd.set_option("mode.chained_assignment", None)


# data frame names
df=None
data=None
#file name 
inputFileName="used_cars_data.csv"
try:
     data = pd.read_csv(inputFileName, index_col="S.No.")
except FileNotFoundError as err:
    print(f'Error : cannot import file : {inputFileName}, Error : {err.args[1]} ')
finally:
    if not isinstance(data,pd.core.frame.DataFrame):
        print(f'Cannot proceed further, Dataframe is null'); 
        #dont want to stop the process in notebook, ideally next steps would not proceed.
    else:
        print(f'input file : {inputFileName} data imported.');
        df = data.copy()

input file : used_cars_data.csv data imported.


# building common methods that we will use multiple times to check data and during data pre processing 

def data_shape(df):
    """
    Prints total rows and columns in input data frame
    """
    print(f"There are {df.shape[0]} rows and {df.shape[1]} columns.")


def print_data_types(df):
    """
    Prints data types, and non null count for all columns in input data frame
    """
    print("\nData Type of each column\n")
    print(df.info())


def print_null_info(df):
    """
    Prints total null value count(s) for all columns in input data frame
    """
    print("\nTotal Null value counts\n")
    print(df.isnull().sum().sort_values(ascending=False))


def print_category_value_counts(df, column_names):
    """
    Prints unique value counts, top 10 value & count(s) for all category columns in input data frame
    """
    print()
    for typeval, col in zip(df[column_names].dtypes, df[column_names]):
            print()
            print(f"Column name : {col} has total {df[col].nunique()} unique values")
            print()
            print(df[col].value_counts()[0:10])
            print()
            print("-" * 50)


df.sample(n=10)


# check data shape, total rows and columns

data_shape(df)

There are 7253 rows and 13 columns.


# check all data types and not null counts
print_data_types(df)

Data Type of each column

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7253 entries, 0 to 7252
Data columns (total 13 columns):
 #   Column             Non-Null Count  Dtype  
---  ------             --------------  -----  
 0   Name               7253 non-null   object 
 1   Location           7253 non-null   object 
 2   Year               7253 non-null   int64  
 3   Kilometers_Driven  7253 non-null   int64  
 4   Fuel_Type          7253 non-null   object 
 5   Transmission       7253 non-null   object 
 6   Owner_Type         7253 non-null   object 
 7   Mileage            7251 non-null   object 
 8   Engine             7207 non-null   object 
 9   Power              7078 non-null   object 
 10  Seats              7200 non-null   float64
 11  New_Price          1006 non-null   object 
 12  Price              6019 non-null   float64
dtypes: float64(2), int64(2), object(9)
memory usage: 793.3+ KB
None


# describe shows summary for all numerical columns, its min, max, mean, 25th,50th and 75th% percentile, st deviation,
df.describe().T


# check how many null values in each columns
print_null_info(df)

Total Null value counts

New_Price            6247
Price                1234
Power                 175
Seats                  53
Engine                 46
Mileage                 2
Name                    0
Location                0
Year                    0
Kilometers_Driven       0
Fuel_Type               0
Transmission            0
Owner_Type              0
dtype: int64


# check for any duplciate data
df[df.duplicated()]


# print value types and value counts
print_category_value_counts(df, ["Name", "Year"])


Column name : Name has total 2041 unique values

Mahindra XUV500 W8 2WD        55
Maruti Swift VDI              49
Maruti Swift Dzire VDI        42
Honda City 1.5 S MT           39
Maruti Swift VDI BSIV         37
Maruti Ritz VDi               35
Toyota Fortuner 3.0 Diesel    35
Hyundai Grand i10 Sportz      32
Honda Brio S MT               32
Honda City 1.5 V MT           32
Name: Name, dtype: int64

--------------------------------------------------

Column name : Year has total 23 unique values

2015    929
2014    925
2016    886
2013    791
2017    709
2012    690
2011    579
2010    407
2018    361
2009    252
Name: Year, dtype: int64

--------------------------------------------------


# print value types and value counts
print_category_value_counts(df, ["Location", "Fuel_Type"])


Column name : Location has total 11 unique values

Mumbai        949
Hyderabad     876
Kochi         772
Coimbatore    772
Pune          765
Delhi         660
Kolkata       654
Chennai       591
Jaipur        499
Bangalore     440
Name: Location, dtype: int64

--------------------------------------------------

Column name : Fuel_Type has total 5 unique values

Diesel      3852
Petrol      3325
CNG           62
LPG           12
Electric       2
Name: Fuel_Type, dtype: int64

--------------------------------------------------


# print value types and value counts
print_category_value_counts(df, ["Transmission", "Owner_Type"])


Column name : Transmission has total 2 unique values

Manual       5204
Automatic    2049
Name: Transmission, dtype: int64

--------------------------------------------------

Column name : Owner_Type has total 4 unique values

First             5952
Second            1152
Third              137
Fourth & Above      12
Name: Owner_Type, dtype: int64

--------------------------------------------------


# print value types and value counts
print_category_value_counts(df, ["Year", "Engine", "Power"])


Column name : Year has total 23 unique values

2015    929
2014    925
2016    886
2013    791
2017    709
2012    690
2011    579
2010    407
2018    361
2009    252
Name: Year, dtype: int64

--------------------------------------------------

Column name : Engine has total 150 unique values

1197 CC    732
1248 CC    610
1498 CC    370
998 CC     309
1198 CC    281
2179 CC    278
1497 CC    273
1968 CC    266
1995 CC    212
1461 CC    188
Name: Engine, dtype: int64

--------------------------------------------------

Column name : Power has total 385 unique values

74 bhp       280
98.6 bhp     166
73.9 bhp     152
140 bhp      142
78.9 bhp     128
67.1 bhp     126
67.04 bhp    125
82 bhp       124
88.5 bhp     120
118 bhp      111
Name: Power, dtype: int64

--------------------------------------------------


df["Seats"].value_counts()

5.000     6047
7.000      796
8.000      170
4.000      119
6.000       38
2.000       18
10.000       8
9.000        3
0.000        1
Name: Seats, dtype: int64


# checking data types
print_data_types(df)

Data Type of each column

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7253 entries, 0 to 7252
Data columns (total 13 columns):
 #   Column             Non-Null Count  Dtype  
---  ------             --------------  -----  
 0   Name               7253 non-null   object 
 1   Location           7253 non-null   object 
 2   Year               7253 non-null   int64  
 3   Kilometers_Driven  7253 non-null   int64  
 4   Fuel_Type          7253 non-null   object 
 5   Transmission       7253 non-null   object 
 6   Owner_Type         7253 non-null   object 
 7   Mileage            7251 non-null   object 
 8   Engine             7207 non-null   object 
 9   Power              7078 non-null   object 
 10  Seats              7200 non-null   float64
 11  New_Price          1006 non-null   object 
 12  Price              6019 non-null   float64
dtypes: float64(2), int64(2), object(9)
memory usage: 793.3+ KB
None


# create a list those columns should be converted as category
cat_vars = ["Location", "Fuel_Type", "Transmission", "Owner_Type"]

for colname in cat_vars:
    df[colname] = df[colname].astype("category")

print_data_types(df)

Data Type of each column

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7253 entries, 0 to 7252
Data columns (total 13 columns):
 #   Column             Non-Null Count  Dtype   
---  ------             --------------  -----   
 0   Name               7253 non-null   object  
 1   Location           7253 non-null   category
 2   Year               7253 non-null   int64   
 3   Kilometers_Driven  7253 non-null   int64   
 4   Fuel_Type          7253 non-null   category
 5   Transmission       7253 non-null   category
 6   Owner_Type         7253 non-null   category
 7   Mileage            7251 non-null   object  
 8   Engine             7207 non-null   object  
 9   Power              7078 non-null   object  
 10  Seats              7200 non-null   float64 
 11  New_Price          1006 non-null   object  
 12  Price              6019 non-null   float64 
dtypes: category(4), float64(2), int64(2), object(5)
memory usage: 595.9+ KB
None


# strip kmpl and km/kg values from Mileage column
df["Mileage"] = df["Mileage"].str.rstrip(" kmpl")
df["Mileage"] = df["Mileage"].str.rstrip(" km/g")

# strip CC value from engine
df["Engine"] = df["Engine"].str.rstrip(" CC")

# strip bhp value from power
df["Power"] = df["Power"].str.rstrip(" bhp")

# create a list those columns should be converted as float
num_vars = ["Mileage", "Engine", "Power"]

for colname in num_vars:
    df[colname] = df[colname].astype("float64")


print_data_types(df)

Data Type of each column

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7253 entries, 0 to 7252
Data columns (total 13 columns):
 #   Column             Non-Null Count  Dtype   
---  ------             --------------  -----   
 0   Name               7253 non-null   object  
 1   Location           7253 non-null   category
 2   Year               7253 non-null   int64   
 3   Kilometers_Driven  7253 non-null   int64   
 4   Fuel_Type          7253 non-null   category
 5   Transmission       7253 non-null   category
 6   Owner_Type         7253 non-null   category
 7   Mileage            7251 non-null   float64 
 8   Engine             7207 non-null   float64 
 9   Power              7078 non-null   float64 
 10  Seats              7200 non-null   float64 
 11  New_Price          1006 non-null   object  
 12  Price              6019 non-null   float64 
dtypes: category(4), float64(5), int64(2), object(2)
memory usage: 595.9+ KB
None


# writing a method to convert price in string to number
def price_to_num(price_val):
    """This function takes in a string representing a price in INR
    and converts it to a number. For example, '9.47 Lakh' becomes 9.47.
    If the input is already numeric, which probably means it's NaN,
    this function just returns np.nan."""
    if isinstance(price_val, str):  # checks if `price_val` is a string
        multiplier = 1  # handles Lakh vs Cr salaries
        if price_val.endswith("Lakh"):
            multiplier = 1
        elif price_val.endswith("Cr"):
            multiplier = 100
        return float((price_val.replace("Lakh", "").replace("Cr", ""))) * multiplier
    else:  # this happens when the current income is np.nan
        return np.nan


df["New_Price"] = df["New_Price"].apply(price_to_num)


# lets check how data looks after concversion 
df.sample(n=10)


# check for data tyes 
print_data_types(df)

Data Type of each column

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7253 entries, 0 to 7252
Data columns (total 13 columns):
 #   Column             Non-Null Count  Dtype   
---  ------             --------------  -----   
 0   Name               7253 non-null   object  
 1   Location           7253 non-null   category
 2   Year               7253 non-null   int64   
 3   Kilometers_Driven  7253 non-null   int64   
 4   Fuel_Type          7253 non-null   category
 5   Transmission       7253 non-null   category
 6   Owner_Type         7253 non-null   category
 7   Mileage            7251 non-null   float64 
 8   Engine             7207 non-null   float64 
 9   Power              7078 non-null   float64 
 10  Seats              7200 non-null   float64 
 11  New_Price          1006 non-null   float64 
 12  Price              6019 non-null   float64 
dtypes: category(4), float64(6), int64(2), object(1)
memory usage: 595.9+ KB
None


# Extract Make and Model From Name.

model_data = pd.DataFrame(
    data=[
        (i, i.split()[0].title(), " ".join(i.split()[1:])) if i == i else (i, "", "")
        for i in df["Name"].str.strip().str.title()
    ],
    columns=["Name", "Make", "Model"],
)


# add new columns to existing data frame

df["Make"] = model_data["Make"]
df["Model"] = model_data["Model"]

# make Make and Model as catrory column

df["Make"] = df["Make"].astype("category")
df["Model"] = df["Model"].astype("category")


# print value types and value counts
print_category_value_counts(model_data, ["Make", "Model"])


Column name : Make has total 32 unique values

Maruti           1444
Hyundai          1340
Honda             743
Toyota            507
Mercedes-Benz     380
Volkswagen        374
Ford              351
Mahindra          331
Bmw               312
Audi              285
Name: Make, dtype: int64

--------------------------------------------------

Column name : Model has total 2017 unique values

Swift Dzire Vdi        62
Xuv500 W8 2Wd          55
Swift Vdi              49
City 1.5 S Mt          39
Swift Vdi Bsiv         37
Swift Dzire Vxi        37
Fortuner 3.0 Diesel    35
Ritz Vdi               35
Alto Lxi               34
Amaze S I-Dtech        32
Name: Model, dtype: int64

--------------------------------------------------


df["Age_Of_Car"] = 2021 - df["Year"]


df.describe().T


## building a Common method to generate graphs


def generate_univariate_chart(data, feature, hue=None, kde=False, bins=20):
    """
    Builds histplot and boxplot for given field.
    Can plot hue, kde and bins based on params, these are optional columns
    """
    sns.set_style("darkgrid")
    print(f"Genrating Charts for feature : {feature}")
    # sns.set_context('poster',font_scale=1)
    # figsize(width,height)

    fig, axes = plt.subplots(2, figsize=(15, 15))
    fig.suptitle("Univariate analysis for " + feature)

    sns.histplot(
        data=data,
        x=feature,
        ax=axes[0],
        palette="winter",
        bins=bins,
        kde=kde,
        hue=hue,
        multiple="dodge",
    )
    sns.boxplot(
        data=data, x=feature, ax=axes[1], showmeans=True, color="violet", hue=hue
    )


# with all params
generate_univariate_chart(data=df, feature="Price", hue=None, bins=10, kde=False)

Genrating Charts for feature : Price


# with all params
generate_univariate_chart(data=df, feature="New_Price", hue=None, bins=10, kde=False)

Genrating Charts for feature : New_Price


# with all params
generate_univariate_chart(
    data=df, feature="Kilometers_Driven", hue=None, bins=10, kde=False
)

Genrating Charts for feature : Kilometers_Driven


df[df["Kilometers_Driven"] >= 6500000].index
df["Kilometers_Driven"][df[df["Kilometers_Driven"] >= 6500000].index] = 65000


# with all params
generate_univariate_chart(
    data=df, feature="Kilometers_Driven", hue=None, bins=10, kde=False
)

Genrating Charts for feature : Kilometers_Driven


# check how many null values in each columns
print_null_info(df)

Total Null value counts

New_Price            6247
Price                1234
Power                 175
Seats                  53
Engine                 46
Mileage                 2
Name                    0
Location                0
Year                    0
Kilometers_Driven       0
Fuel_Type               0
Transmission            0
Owner_Type              0
Make                    0
Model                   0
Age_Of_Car              0
dtype: int64


# most rows don't have missing values now
num_missing = df.isnull().sum(axis=1)
num_missing.value_counts()

1    5232
2    1125
0     823
4      36
3      27
5      10
dtype: int64


df[num_missing == 4].sample(n=10)


## to find data from exiting
def find_data_for_engine(name, year):
    """
    find data from existing data frame, using input name and year where engine is not null
    """
    idx = np.where((df["Name"] == name) & (df["Year"] == year) & df["Engine"].notna())
    return idx


missing_items = df[df["Engine"].isna()][["Name", "Year"]]

print(f"Missing values before fix : {missing_items.size}")

for ind in missing_items.index:
    missing_name = missing_items["Name"][ind]
    missing_year = missing_items["Year"][ind]
    value_index = find_data_for_engine(missing_name, missing_year)
    if len(value_index[0]) > 0:
        value = df["Engine"][value_index[0][0]]
        df["Engine"][ind] = value

## Fill median values by Year and Make

df["Engine"] = df.groupby(["Year", "Make"])["Engine"].apply(
    lambda x: x.fillna(x.median())
)

## Fill median values by Make

df["Engine"] = df.groupby(["Make"])["Engine"].apply(lambda x: x.fillna(x.median()))

missing_items = df[df["Engine"].isna()][["Name", "Year"]]
print(f"Missing values after fix : {missing_items.size}")

Missing values before fix : 92
Missing values after fix : 0


## to find data from exiting
def find_data_for_power(name, year):
    """
    find data from existing data frame, using input name and year where power is not null
    """
    idx = np.where((df["Name"] == name) & (df["Year"] == year) & df["Power"].notna())
    return idx


missing_items = df[df["Power"].isna()][["Name", "Year"]]

print(f"Missing values before fix : {missing_items.size}")

for ind in missing_items.index:
    missing_name = missing_items["Name"][ind]
    missing_year = missing_items["Year"][ind]
    value_index = find_data_for_power(missing_name, missing_year)
    if len(value_index[0]) > 0:
        value = df["Power"][value_index[0][0]]
        df["Power"][ind] = value

## Fill median values by Year and Make
df["Power"] = df.groupby(["Year", "Make"])["Power"].apply(
    lambda x: x.fillna(x.median())
)
## Fill median values by Make
df["Power"] = df.groupby(["Make"])["Power"].apply(lambda x: x.fillna(x.median()))

missing_items = df[df["Power"].isna()][["Name", "Year"]]
print(f"Missing values after fix : {missing_items.size}")

Missing values before fix : 350
Missing values after fix : 4


## to find data from exiting
def find_data_for_seat(name, year):
    """
    find data from existing data frame, using input name and year where power is not null
    """
    idx = np.where((df["Name"] == name) & (df["Year"] == year) & df["Seats"].notna())
    return idx


missing_items = df[df["Seats"].isna()][["Name", "Year"]]

print(f"Missing values before fix : {missing_items.size}")

for ind in missing_items.index:
    missing_name = missing_items["Name"][ind]
    missing_year = missing_items["Year"][ind]
    value_index = find_data_for_seat(missing_name, missing_year)
    if len(value_index[0]) > 0:
        value = df["Seats"][value_index[0][0]]
        df["Seats"][ind] = value

## Fill median values by Year and Make
df["Seats"] = df.groupby(["Year", "Make"])["Seats"].apply(
    lambda x: x.fillna(x.median())
)
## Fill median values by Make
df["Seats"] = df.groupby(["Make"])["Seats"].apply(lambda x: x.fillna(x.median()))

missing_items = df[df["Seats"].isna()][["Name", "Year"]]
print(f"Missing values after fix : {missing_items.size}")

Missing values before fix : 106
Missing values after fix : 0


# check how many null values in each columns
print_null_info(df)

Total Null value counts

New_Price            6247
Price                1234
Mileage                 2
Power                   2
Name                    0
Location                0
Year                    0
Kilometers_Driven       0
Fuel_Type               0
Transmission            0
Owner_Type              0
Engine                  0
Seats                   0
Make                    0
Model                   0
Age_Of_Car              0
dtype: int64


missing_items = df[df["Price"].isna()][["Name"]]
print(f"Missing values before fix : {missing_items.size}")


## Fill median values by Year, Make and Model
df["Price"] = df.groupby(["Year", "Make", "Model"])["Price"].apply(
    lambda x: x.fillna(x.median())
)
## Fill median values by Year, Make
df["Price"] = df.groupby(["Year", "Make"])["Price"].apply(
    lambda x: x.fillna(x.median())
)

missing_items = df[df["Price"].isna()][["Name"]]
print(f"Missing values after fix : {missing_items.size}")

Missing values before fix : 12
Missing values after fix : 12


missing_items = df[df["New_Price"].isna()][["Name"]]
print(f"Missing values after fix : {missing_items.size}")


## Fill median values by Year, Make and Model
df["New_Price"] = df.groupby(["Year", "Make", "Model"])["New_Price"].apply(
    lambda x: x.fillna(x.median())
)
## Fill median values by Year, Make
df["New_Price"] = df.groupby(["Year", "Make"])["New_Price"].apply(
    lambda x: x.fillna(x.median())
)

## Fill median values by Make, Model
df["New_Price"] = df.groupby(["Make", "Model"])["New_Price"].apply(
    lambda x: x.fillna(x.median())
)


missing_items = df[df["New_Price"].isna()][["Name"]]
print(f"Missing values after fix : {missing_items.size}")

Missing values after fix : 224
Missing values after fix : 224


# check how many null values in each columns
print_null_info(df)

Total Null value counts

New_Price            224
Price                 12
Mileage                2
Power                  2
Name                   0
Location               0
Year                   0
Kilometers_Driven      0
Fuel_Type              0
Transmission           0
Owner_Type             0
Engine                 0
Seats                  0
Make                   0
Model                  0
Age_Of_Car             0
dtype: int64


df.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7253 entries, 0 to 7252
Data columns (total 16 columns):
 #   Column             Non-Null Count  Dtype   
---  ------             --------------  -----   
 0   Name               7253 non-null   object  
 1   Location           7253 non-null   category
 2   Year               7253 non-null   int64   
 3   Kilometers_Driven  7253 non-null   int64   
 4   Fuel_Type          7253 non-null   category
 5   Transmission       7253 non-null   category
 6   Owner_Type         7253 non-null   category
 7   Mileage            7251 non-null   float64 
 8   Engine             7253 non-null   float64 
 9   Power              7251 non-null   float64 
 10  Seats              7253 non-null   float64 
 11  New_Price          7029 non-null   float64 
 12  Price              7241 non-null   float64 
 13  Make               7253 non-null   category
 14  Model              7253 non-null   category
 15  Age_Of_Car         7253 non-null   int64   
dtypes: category(6), float64(6), int64(3), object(1)
memory usage: 1013.4+ KB


# writing a method that can take a column as input and print all the values falling outside the range of Limts of IQR 
def check_outlier_using_IQR(column, limit=10):
    """
    Check Limts * IQR for each values and prints the ones falls outside the range.
    """
    quartiles = np.quantile(df[column][df[column].notnull()], [0.25, 0.75])
    limit_iqr = limit * (quartiles[1] - quartiles[0])
    outlier = df.loc[np.abs(df[column] - df[column].median()) > limit_iqr, column]
    print();
    print(f"Column : {column} Outlier(s) check");
    print(f"Mean : {df[column].mean()}, Median : {df[column].median()}, Min : {df[column].min()}, Max : {df[column].max()}");
    print(f"Q1 = {quartiles[0]}, Q3 = {quartiles[1]}, {limit}*IQR = {limit_iqr}, Total Outlier(s) : {outlier.size} \n")
    print(outlier)
    print("-" * 50)


for colname in df.columns[df.dtypes == "float64"]:  # filter float64 columns
    check_outlier_using_IQR(colname)
for colname in df.columns[df.dtypes == "int64"]:  # filter int64 columns
    check_outlier_using_IQR(colname)

Column : Mileage Outlier(s) check
Mean : 18.14158047165911, Median : 18.16, Min : 0.0, Max : 33.54
Q1 = 15.17, Q3 = 21.1, 10*IQR = 59.30000000000001, Total Outlier(s) : 0 

Series([], Name: Mileage, dtype: float64)
--------------------------------------------------

Column : Engine Outlier(s) check
Mean : 1616.3930787260445, Median : 1493.0, Min : 72.0, Max : 5998.0
Q1 = 1198.0, Q3 = 1968.0, 10*IQR = 7700.0, Total Outlier(s) : 0 

Series([], Name: Engine, dtype: float64)
--------------------------------------------------

Column : Power Outlier(s) check
Mean : 112.40513791201128, Median : 93.7, Min : 34.2, Max : 616.0
Q1 = 75.0, Q3 = 138.03, 10*IQR = 630.3, Total Outlier(s) : 0 

Series([], Name: Power, dtype: float64)
--------------------------------------------------

Column : Seats Outlier(s) check
Mean : 5.278643320005515, Median : 5.0, Min : 0.0, Max : 10.0
Q1 = 5.0, Q3 = 5.0, 10*IQR = 0.0, Total Outlier(s) : 1158 

S.No.
3      7.000
7      8.000
15     7.000
29     7.000
32     7.000
        ... 
7175   7.000
7194   4.000
7198   7.000
7206   7.000
7223   7.000
Name: Seats, Length: 1158, dtype: float64
--------------------------------------------------

Column : New_Price Outlier(s) check
Mean : 22.029273100725643, Median : 11.15, Min : 3.91, Max : 375.0
Q1 = 8.254999999999999, Q3 = 19.27, 10*IQR = 110.15, Total Outlier(s) : 187 

S.No.
4      128.000
13     139.000
14     139.000
63     171.000
126    122.977
148    128.000
215    171.000
216    128.000
311    127.000
404    127.000
410    139.000
426    171.000
480    160.000
506    124.880
521    127.000
525    122.977
536    171.000
557    128.000
589    129.285
645    171.000
701    139.000
704    128.000
717    128.000
755    171.000
821    171.000
822    128.000
887    171.000
902    128.000
932    171.000
986    128.000
1038   139.000
1090   128.000
1134   171.000
1204   139.000
1237   128.000
1265   128.000
1299   139.000
1316   128.000
1352   139.000
1356   171.000
1446   171.000
1477   122.977
1502   171.000
1505   139.000
1513   139.000
1570   171.000
1626   171.000
1654   171.000
1766   171.000
1868   128.000
1916   171.000
1918   128.000
1930   171.000
1988   139.000
1999   171.000
2005   128.000
2024   128.000
2040   171.000
2044   128.000
2056   140.000
2084   128.000
2178   127.000
2201   122.977
2263   139.000
2284   128.000
2411   171.000
2529   122.977
2690   128.000
2863   128.000
2884   171.000
2904   171.000
2929   128.000
3085   128.000
3132   136.000
3162   128.000
3199   166.000
3218   160.000
3236   171.000
3242   139.000
3261   128.000
3328   122.977
3410   139.000
3521   124.880
3604   128.000
3607   171.000
3621   122.977
3724   122.977
3752   160.000
3770   122.977
3839   139.000
3899   128.000
3939   171.000
4021   171.000
4039   171.000
4061   128.000
4079   230.000
4105   128.000
4136   122.977
4143   171.000
4153   171.000
4245   171.000
4247   139.000
4261   128.000
4267   122.977
4292   171.000
4311   122.977
4342   129.285
4466   139.000
4517   128.000
4581   128.000
4592   171.000
4602   171.000
4630   139.000
4674   171.000
4687   139.000
4701   139.000
4714   171.000
4757   122.977
4778   171.000
4801   139.000
4813   128.000
4814   139.000
4833   122.977
4846   128.000
4868   128.000
4869   139.000
4922   128.000
4925   127.000
5016   139.000
5022   139.000
5075   171.000
5142   171.000
5153   128.000
5239   128.000
5253   171.000
5282   171.000
5305   171.000
5311   139.000
5346   128.000
5382   160.000
5410   171.000
5411   139.000
5414   171.000
5416   139.000
5461   124.880
5530   139.000
5545   139.000
5586   128.000
5686   171.000
5728   122.977
5741   128.000
5842   171.000
5874   171.000
5885   171.000
6001   128.000
6006   122.977
6029   128.000
6060   128.000
6061   128.000
6072   128.000
6204   128.000
6208   128.000
6212   158.000
6287   128.000
6296   171.000
6354   375.000
6367   160.000
6383   122.977
6399   128.000
6413   128.000
6417   128.000
6426   160.000
6434   139.000
6458   171.000
6473   171.000
6513   128.000
6579   128.000
6704   171.000
6857   139.000
6951   171.000
6986   171.000
6989   171.000
7007   171.000
7072   128.000
7157   160.000
7180   171.000
7198   139.000
Name: New_Price, dtype: float64
--------------------------------------------------

Column : Price Outlier(s) check
Mean : 9.35116903742574, Median : 5.63, Min : 0.44, Max : 160.0
Q1 = 3.5, Q3 = 9.87, 10*IQR = 63.69999999999999, Total Outlier(s) : 26 

S.No.
282     70.360
418     70.990
655     75.000
1194    69.500
1352    70.000
1505    97.070
1583    72.940
1690    70.800
1885    79.000
1974    93.670
1984    93.000
2095    83.960
2319    78.800
2422    79.000
2541    70.150
2659    75.000
2907    75.000
3489    75.000
3752    70.660
4079   160.000
4614    70.000
4691    90.000
5535    85.000
5603    72.000
5781   120.000
5919   100.000
Name: Price, dtype: float64
--------------------------------------------------

Column : Year Outlier(s) check
Mean : 2013.3653660554253, Median : 2014.0, Min : 1996, Max : 2019
Q1 = 2011.0, Q3 = 2016.0, 10*IQR = 50.0, Total Outlier(s) : 0 

Series([], Name: Year, dtype: int64)
--------------------------------------------------

Column : Kilometers_Driven Outlier(s) check
Mean : 57811.84406452502, Median : 53416.0, Min : 171, Max : 775000
Q1 = 34000.0, Q3 = 73000.0, 10*IQR = 390000.0, Total Outlier(s) : 6 

S.No.
340     775000
358     620000
1860    720000
2823    480000
3092    480000
4491    445000
Name: Kilometers_Driven, dtype: int64
--------------------------------------------------

Column : Age_Of_Car Outlier(s) check
Mean : 7.6346339445746585, Median : 7.0, Min : 2, Max : 25
Q1 = 5.0, Q3 = 10.0, 10*IQR = 50.0, Total Outlier(s) : 0 

Series([], Name: Age_Of_Car, dtype: int64)
--------------------------------------------------


cols_to_log = ["Price", "New_Price", "Kilometers_Driven"]
for colname in cols_to_log:
    plt.hist(df[colname], bins=50)
    plt.title(colname + " without log")
    plt.show()
    plt.hist(np.log(df[colname]), 50)
    plt.title(colname + " with log")
    plt.show()


cols_to_log = ["Price", "New_Price", "Kilometers_Driven"]
for colname in cols_to_log:
    df[colname + "_log"] = np.log(df[colname])
df.drop(cols_to_log, axis=1, inplace=True)


# check how many null values in each columns
print_null_info(df)

Total Null value counts

New_Price_log            224
Price_log                 12
Mileage                    2
Power                      2
Name                       0
Location                   0
Year                       0
Fuel_Type                  0
Transmission               0
Owner_Type                 0
Engine                     0
Seats                      0
Make                       0
Model                      0
Age_Of_Car                 0
Kilometers_Driven_log      0
dtype: int64


# most rows don't have missing values now
num_missing = df.isnull().sum(axis=1)
num_missing.value_counts()

0    7019
1     229
2       4
3       1
dtype: int64


df.dropna(inplace=True)


# check how many null values in each columns
print_null_info(df)

Total Null value counts

Name                     0
Location                 0
Year                     0
Fuel_Type                0
Transmission             0
Owner_Type               0
Mileage                  0
Engine                   0
Power                    0
Seats                    0
Make                     0
Model                    0
Age_Of_Car               0
Price_log                0
New_Price_log            0
Kilometers_Driven_log    0
dtype: int64


# Price Ranges
df["Price_Type"] = binned_price = pd.cut(
    df["Price_log"],
    [-np.inf, 1.5, 3, np.inf],
    labels=["ECONOMY", "MID-SCALE", "LUXURY"],
)


df.sample(n=30)


df.drop(["Name", "Year"], axis=1, inplace=True)


df.sample(n=10)


# Let's look at the statistical summary of the data
df.describe().T


#we have already created common chart method for univariate analysis


#generating univariate chart for price column to check data spread and outlier range
generate_univariate_chart(data=df, feature="Price_log", hue=None, bins=10, kde=False)

Genrating Charts for feature : Price_log


## Finding mean value - using exp

np.exp(df["Price_log"].mean())

6.295193239506662


#generating univariate chart for new price column to check data spread and outlier range
generate_univariate_chart(
    data=df, feature="New_Price_log", hue=None, bins=10, kde=False
)

Genrating Charts for feature : New_Price_log


## EXPanding Log 3 and
print(f"Majority of new price are under {np.exp(3)}")

##outliers
print(f"new price outliers are beyond {np.exp(5)}")

Majority of new price are under 20.085536923187668
new price outliers are beyond 148.4131591025766


#generating univariate chart for age of car column to check data spread and outlier range
generate_univariate_chart(data=df, feature="Age_Of_Car", hue=None, bins=10, kde=False)

Genrating Charts for feature : Age_Of_Car


#generating univariate chart for mileage column to check data spread and outlier range
generate_univariate_chart(data=df, feature="Mileage", hue=None, bins=10, kde=False)

Genrating Charts for feature : Mileage


#generating univariate chart for engine size column to check data spread and outlier range
generate_univariate_chart(data=df, feature="Engine", hue=None, bins=10, kde=False)

Genrating Charts for feature : Engine


# generating univariate chart for power column to check data spread and outlier range
generate_univariate_chart(data=df, feature="Power", hue=None, bins=10, kde=False)

Genrating Charts for feature : Power


# generating univariate chart for price column to check data spread and outlier range
generate_univariate_chart(data=df, feature="Seats", hue=None, bins=10, kde=False)

# converting seats to category type column

df["Seats"] = df["Seats"].astype("float64")

Genrating Charts for feature : Seats


# with all params
generate_univariate_chart(
    data=df, feature="Kilometers_Driven_log", hue=None, bins=10, kde=False
)

Genrating Charts for feature : Kilometers_Driven_log


## Common method to genrate count graphs for category type columns


def generate_count_chart(df, feature):
    """
    Builds count plot for given column
    """
    plt.figure(figsize=(20, 8))
    chart = sns.countplot(x=feature, data=df, order=df[feature].value_counts().index)
    chart.set_xticklabels(
        chart.get_xticklabels(),
        rotation=45,
        horizontalalignment="right",
        fontweight="light",
        fontsize="x-large",
    )


generate_count_chart(df, "Make")


generate_count_chart(df, "Price_Type")


generate_count_chart(df, "Age_Of_Car")


generate_count_chart(df, "Location")


generate_count_chart(df, "Fuel_Type")


generate_count_chart(df, "Transmission")


generate_count_chart(df, "Owner_Type")


generate_count_chart(df, "Seats")


# Generate Corelation chart and understand how these features are related
def generate_matrix_chart(data):
    """
    Generate corelation chart(heat-map) between all numerial features 
    """
    sns.set_style("darkgrid")
    plt.figure(figsize=(12, 10))
    sns.heatmap(
        data.corr(), cmap="Spectral", linecolor="white", linewidths=1, annot=True
    )


generate_matrix_chart(df)


def generate_pairplot_chart(data, hue=None):
    """
    Generate pair plot for all features
    """
    sns.set_style("darkgrid")
    sns.pairplot(data, hue=hue, palette="magma")


generate_pairplot_chart(df)


## this method generate joint plot of x vs y feature
def generate_bivariate_chart(data, xfeature, yfeature, hue=None):
    """
    common method to generate joint plot for various columns
    hue param is optional
    """
    sns.set_style("darkgrid")
    print(f"Genrating Charts for feature x : {xfeature}, y : {yfeature}")

    sns.jointplot(
        data=data,
        x=xfeature,
        y=yfeature,
        palette="winter",
        height=10,
        kind="scatter",
        hue=hue,
    )


generate_bivariate_chart(
    xfeature="Power", yfeature="Price_log", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Power, y : Price_log


generate_bivariate_chart(
    xfeature="Engine", yfeature="Price_log", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Engine, y : Price_log


generate_bivariate_chart(
    xfeature="New_Price_log", yfeature="Price_log", data=df, hue="Price_Type"
)

Genrating Charts for feature x : New_Price_log, y : Price_log


generate_bivariate_chart(
    xfeature="Mileage", yfeature="Engine", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Mileage, y : Engine


generate_bivariate_chart(
    xfeature="Mileage", yfeature="Power", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Mileage, y : Power


generate_bivariate_chart(xfeature="Engine", yfeature="Power", data=df, hue="Price_Type")

Genrating Charts for feature x : Engine, y : Power


generate_bivariate_chart(
    xfeature="Power", yfeature="New_Price_log", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Power, y : New_Price_log


generate_bivariate_chart(
    xfeature="Engine", yfeature="New_Price_log", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Engine, y : New_Price_log


# method to generate x vs y count plot and box plot
def generate_category_chart(data, xfeature, yfeature, hue=None):
    """
    generate box plot and count plot for given x and y 
    """
    sns.set_style("darkgrid")
    print(f"Genrating Charts for feature x : {xfeature}, y : {yfeature}")

    # sns.jointplot(data=data , x=xfeature, y=yfeature, palette="winter", height=8,kind='scatter', hue=hue);
    fig, axes = plt.subplots(2, figsize=(15, 20))
    fig.suptitle(xfeature + " vs " + yfeature)

    chart = sns.countplot(
        x=xfeature,
        data=data,
        ax=axes[0],
        hue=hue,
        order=df[xfeature].value_counts().index,
    )
    chart.set_xticklabels(
        chart.get_xticklabels(),
        rotation=45,
        horizontalalignment="right",
        fontweight="light",
        fontsize="x-large",
    )

    sns.boxplot(x=xfeature, y=yfeature, hue=hue, data=data, ax=axes[1])
    plt.xticks(rotation=90)


generate_category_chart(
    xfeature="Location", yfeature="Price_log", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Location, y : Price_log


generate_category_chart(
    xfeature="Fuel_Type", yfeature="Price_log", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Fuel_Type, y : Price_log


generate_category_chart(
    xfeature="Transmission", yfeature="Price_log", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Transmission, y : Price_log


generate_category_chart(
    xfeature="Owner_Type", yfeature="Price_log", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Owner_Type, y : Price_log


generate_category_chart(
    xfeature="Make", yfeature="Price_log", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Make, y : Price_log


generate_category_chart(
    xfeature="Age_Of_Car", yfeature="Price_log", data=df, hue="Price_Type"
)

Genrating Charts for feature x : Age_Of_Car, y : Price_log


# we dont need Model, Delete Model Column(its almost unique)

X_1 = df.drop(["Price_log", "Model"], axis=1)
y_1 = df["Price_log"]

X_1.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7019 entries, 0 to 7252
Data columns (total 13 columns):
 #   Column                 Non-Null Count  Dtype   
---  ------                 --------------  -----   
 0   Location               7019 non-null   category
 1   Fuel_Type              7019 non-null   category
 2   Transmission           7019 non-null   category
 3   Owner_Type             7019 non-null   category
 4   Mileage                7019 non-null   float64 
 5   Engine                 7019 non-null   float64 
 6   Power                  7019 non-null   float64 
 7   Seats                  7019 non-null   float64 
 8   Make                   7019 non-null   category
 9   Age_Of_Car             7019 non-null   int64   
 10  New_Price_log          7019 non-null   float64 
 11  Kilometers_Driven_log  7019 non-null   float64 
 12  Price_Type             7019 non-null   category
dtypes: category(6), float64(6), int64(1)
memory usage: 740.2 KB


X_1 = pd.get_dummies(
    X_1,
    columns=X_1.select_dtypes(include=["object", "category"]).columns.tolist(),
    drop_first=True,
)
X_1.head()


X_1.shape

(7019, 58)


# splitting the data in 70:30 ratio for train to test data, Random seed = 20

x_train_1, x_test_1, y_train_1, y_test_1 = train_test_split(
    X_1, y_1, test_size=0.3, random_state=20
)


print("Total rows in train :", x_train.shape[0])
print("Total rows in test :", x_test.shape[0])


## buidling a LinearRegression and fit it using training set
# and creating a name label to track what type of model it is

lm_1 = LinearRegression()
lm_1.fit(x_train_1, y_train_1)
lm_1_name = "With all Features"


# function to compute adjusted R-squared
def adj_r2_score(predictors, targets, predictions):
    r2 = r2_score(targets, predictions)
    n = predictors.shape[0]
    k = predictors.shape[1]
    return 1 - ((1 - r2) * (n - 1) / (n - k - 1))


# function to compute different metrics to check performance of a regression model
def model_performance_regression(
    model, predictors, target, model_type=None, run_type=None
):
    """
    Function to compute different metrics to check regression model performance

    model: regressor
    predictors: independent variables
    target: dependent variable
    """

    # predicting using the independent variables
    pred = model.predict(predictors)

    r2 = r2_score(target, pred)  # to compute R-squared
    adjr2 = adj_r2_score(predictors, target, pred)  # to compute adjusted R-squared
    rmse = np.sqrt(mean_squared_error(target, pred))  # to compute RMSE
    mae = mean_absolute_error(target, pred)  # to compute MAE
    mse = mean_squared_error(target, pred)  # to compute MAE

    y_mean = target.mean()
    SST = ((target - y_mean) ** 2).sum()
    SSE = ((target - model.predict(predictors)) ** 2).sum()
    # print(f"R-squared : {r2}")
    # print(f"Adj. R-squared : {adjr2}")
    # print(f"RMSE : {rmse}")
    # print(f"MAE : {mae}")
    # print(f"MSE : {mse}")
    # print(f"SST : {SST}")
    # print(f"SSE : {SSE}")

    # creating a dataframe of metrics
    df_perf = pd.DataFrame(
        {
            "Model": model_type,
            "Run Type": run_type,
            "Total Features": len(predictors.columns),
            "Size": target.size,
            "score %": (r2 * 100),
            "R-squared": r2,
            "Adj. R-squared": adjr2,
            "RMSE": rmse,
            "MAE": mae,
            "MSE": mse,
            "SST": SST,
            "SSE": SSE,
        },
        index=[0],
    )

    return df_perf


# Checking model performance on train set
# print("Training Performance\n")

perf_training_1 = model_performance_regression(
    lm_1, x_train_1, y_train_1, lm_1_name, "Training"
)


# Checking model performance on test set
# print("Test Performance\n")
perf_testing_1 = model_performance_regression(
    lm_1, x_test_1, y_test_1, lm_1_name, "Test"
)
perf_training_1_result = pd.concat([perf_training_1, perf_testing_1])
perf_training_1_result


# creating a new data frame for this model so that existing data wont be impacted
df_2 = df.copy()


## Deleting CNG,LPG,Electric

indexNames = df_2[ (df_2['Fuel_Type'] == "CNG") | (df_2['Fuel_Type'] == "LPG")  | (df_2['Fuel_Type'] == "Electric") ].index

# Delete these row indexes from dataFrame
df_2.drop(indexNames , inplace=True)


## Deleting seats not 5 or 7

indexNames = df_2[ (df_2['Seats'] != 5) & (df_2['Seats'] !=7)   ].index

# Delete these row indexes from dataFrame
df_2.drop(indexNames , inplace=True)

## Deleting owner not first or second

indexNames = df_2[ (df_2['Owner_Type'] != "First") & (df_2['Owner_Type'] !="Second")   ].index

# Delete these row indexes from dataFrame
df_2.drop(indexNames , inplace=True)


# we dont need Model, Delete Model Column(its almost unique)

X_2 = df_2.drop(["Price_log", "Model"], axis=1)
y_2 = df_2["Price_log"]


X_2 = pd.get_dummies(
    X_2,
    columns=X_2.select_dtypes(include=["object", "category"]).columns.tolist(),
    drop_first=True,
)
X_2.head()


# splitting the data in 70:30 ratio for train to test data, Random seed = 20

x_train_2, x_test_2, y_train_2, y_test_2 = train_test_split(
    X_2, y_2, test_size=0.3, random_state=20
)


## buidling a LinearRegression and fit it using training set
# and creating a name label to track what type of model it is

lm_2 = LinearRegression()
lm_2.fit(x_train_2, y_train_2)
lm_2_name = "With all Features with higly sale category"


# Checking model performance on train set
# print("Training Performance\n")

perf_training_2 = model_performance_regression(
    lm_2, x_train_2, y_train_2, lm_2_name, "Training"
)


# Checking model performance on test set
# print("Test Performance\n")
perf_testing_2 = model_performance_regression(
    lm_2, x_test_2, y_test_2, lm_2_name, "Test"
)
perf_training_2_result = pd.concat([perf_training_2, perf_testing_2])
perf_training_2_result


perf_training_compare = pd.concat([perf_training_1_result, perf_training_2_result])
perf_training_compare


# we dont need Model, Delete Model Column(its almost unique)
df_3 = df_2.copy()
X_3 = df_3.drop(["Price_log", "Model", "Power", "New_Price_log"], axis=1)
y_3 = df_3["Price_log"]


X_3 = pd.get_dummies(
    X_3,
    columns=X_3.select_dtypes(include=["object", "category"]).columns.tolist(),
    drop_first=True,
)



# splitting the data in 70:30 ratio for train to test data, Random seed = 20

x_train_3, x_test_3, y_train_3, y_test_3 = train_test_split(
    X_3, y_3, test_size=0.3, random_state=20
)


## buidling a LinearRegression and fit it using training set
# and creating a name label to track what type of model it is

lm_3 = LinearRegression()
lm_3.fit(x_train_3, y_train_3)
lm_3_name = "With Model 2 Featuresm, No Power, No New Price"




# Checking model performance on train set
# print("Training Performance\n")

perf_training_3 = model_performance_regression(
    lm_3, x_train_3, y_train_3, lm_3_name, "Training"
)


# Checking model performance on test set
# print("Test Performance\n")
perf_testing_3 = model_performance_regression(
    lm_3, x_test_3, y_test_3, lm_3_name, "Test"
)
perf_training_3_result = pd.concat([perf_training_3, perf_testing_3])
perf_training_3_result


perf_training_compare = pd.concat(
    [perf_training_1_result, perf_training_2_result, perf_training_3_result]
)
perf_training_compare


df.corr().T


# we dont need Model, Delete Model Column(its almost unique)
df_4 = df.copy()
X_4 = df_4.drop(
    ["Price_log", "Model", "Mileage", "Seats", "Kilometers_Driven_log"], axis=1
)
y_4 = df_4["Price_log"]


X_4 = pd.get_dummies(
    X_4,
    columns=X_4.select_dtypes(include=["object", "category"]).columns.tolist(),
    drop_first=True,
)


# splitting the data in 70:30 ratio for train to test data, Random seed = 20

x_train_4, x_test_4, y_train_4, y_test_4 = train_test_split(
    X_4, y_4, test_size=0.3, random_state=20
)


## buidling a LinearRegression and fit it using training set
# and creating a name label to track what type of model it is

lm_4 = LinearRegression()
lm_4.fit(x_train_4, y_train_4)
lm_4_name = "With Highly Correlated Features"


# Checking model performance on train set
# print("Training Performance\n")

perf_training_4 = model_performance_regression(
    lm_4, x_train_4, y_train_4, lm_4_name, "Training"
)


# Checking model performance on test set
# print("Test Performance\n")
perf_testing_4 = model_performance_regression(
    lm_4, x_test_4, y_test_4, lm_4_name, "Test"
)
perf_training_4_result = pd.concat([perf_training_4, perf_testing_4])
perf_training_4_result


perf_training_compare = pd.concat(
    [
        perf_training_1_result,
        perf_training_2_result,
        perf_training_3_result,
        perf_training_4_result,
    ]
)
perf_training_compare


from mlxtend.feature_selection import SequentialFeatureSelector as SFS


# prepare data

# we dont need Model, Delete Model Column(its almost unique)
df_5 = df.copy()
X_5 = df_5.drop(["Price_log", "Model"], axis=1)
y_5 = df_5["Price_log"]


X_5 = pd.get_dummies(
    X_5,
    columns=X_5.select_dtypes(include=["object", "category"]).columns.tolist(),
    drop_first=True,
)


# splitting the data in 70:30 ratio for train to test data, Random seed = 20

x_train_5, x_test_5, y_train_5, y_test_5 = train_test_split(
    X_5, y_5, test_size=0.3, random_state=20
)


reg = LinearRegression()

# Build step forward feature selection
sfs = SFS(
    reg,
    k_features=x_train.shape[1],
    forward=True,  # k_features denotes the number of features to select
    floating=False,
    n_jobs=-1,
    scoring="r2",
    cv=5,
)


# Perform SFFS
sfs = sfs.fit(x_train_5, y_train_5)


# to plot the performance with addition of each feature
from mlxtend.plotting import plot_sequential_feature_selection as plot_sfs

fig1 = plot_sfs(sfs.get_metric_dict(), kind="std_err", figsize=(15, 5))
plt.title("Sequential Forward Selection (w. StdErr)")
plt.xticks(rotation=90)
plt.show()


reg = LinearRegression()

# Build step forward feature selection
sfs = SFS(
    reg, k_features=28, forward=True, floating=False, scoring="r2", n_jobs=-1, cv=5,
)

# Perform SFFS
sfs = sfs.fit(x_train_5, y_train_5)


# let us select the features which are important
feat_cols = list(sfs.k_feature_idx_)
print(feat_cols)

[0, 1, 2, 3, 4, 5, 6, 7, 9, 11, 14, 17, 21, 25, 27, 37, 38, 40, 42, 43, 44, 45, 50, 52, 53, 55, 56, 57]


# let us look at the names of the important features
x_train_5.columns[feat_cols]

Index(['Mileage', 'Engine', 'Power', 'Seats', 'Age_Of_Car', 'New_Price_log',
       'Kilometers_Driven_log', 'Location_Bangalore', 'Location_Coimbatore',
       'Location_Hyderabad', 'Location_Kolkata', 'Fuel_Type_Diesel',
       'Transmission_Manual', 'Make_Audi', 'Make_Bmw', 'Make_Jaguar',
       'Make_Jeep', 'Make_Land', 'Make_Maruti', 'Make_Mercedes-Benz',
       'Make_Mini', 'Make_Mitsubishi', 'Make_Skoda', 'Make_Tata',
       'Make_Toyota', 'Make_Volvo', 'Price_Type_MID-SCALE',
       'Price_Type_LUXURY'],
      dtype='object')


x_train_final = x_train_5[x_train_5.columns[feat_cols]]
# Creating new x_test with the same variables that we selected for x_train
x_test_final = x_test_5[x_train_final.columns]


# Fitting linear model
lin_reg_model2 = LinearRegression()
lin_reg_model2.fit(x_train_final, y_train_5)

LinearRegression()


# Checking model performance on train set
print("Training Performance\n")
lin_reg_model_train_perf = model_performance_regression(
    lin_reg_model2, x_train_final, y_train_5, "Top 28 Features", "Train"
)
lin_reg_model_train_perf

Training Performance


# Checking model performance on test set
print("Test Performance\n")
lin_reg_model_test_perf = model_performance_regression(
    lin_reg_model2, x_test_final, y_test_5, "Top 28 Features", "Test"
)
lin_reg_model_test_perf

Test Performance


perf_training_compare = pd.concat(
    [
        perf_training_1_result,
        perf_training_2_result,
        perf_training_3_result,
        perf_training_4_result,
        lin_reg_model_train_perf,
        lin_reg_model_test_perf,
    ]
)
perf_training_compare


predictions = lm_2.predict( x_test_2)
plt.figure(figsize=(12,10))
plt.scatter(y_test_2,predictions)
plt.xlabel('Y Test')
plt.ylabel('Predicted Y')

Text(0, 0.5, 'Predicted Y')


plt.figure(figsize=(12, 10))
sns.histplot((y_test_2 - predictions), bins=10, kde=True)

<AxesSubplot:xlabel='Price_log', ylabel='Count'>


# let's check the coefficients and intercept of the model

coef_df = pd.DataFrame(
    np.append(lm_2.intercept_, lm.coef_),
    index=["Intercept"] + x_train_2.columns.tolist(),
    columns=["Coefficients"],
)

coef_df


## key Features to get high prices 

coef_df[coef_df > 0.2].dropna(axis=0)


## key Features to get high prices

coef_df[coef_df < -0.05].dropna(axis=0)


# Let us write the equation of linear regression

Equation = "Price = " + str(lm.intercept_)
print(Equation, end=" ")

for i in range(len(x_train_2.columns)):
    if i != len(x_train_2.columns) - 1:
        print(
            "\n+ (", lm_2.coef_[i], ")*(", x_train_2.columns[i], ")", end="  ",
        )
    else:
        print("\n+ (", lm_2.coef_[i], ")*(", x_train_2.columns[i], ")")

Price = 1.9675703833043392 
+ ( -0.0091536066246779 )*( Mileage )  
+ ( 8.800011588209187e-05 )*( Engine )  
+ ( 0.0044522015922807134 )*( Power )  
+ ( 0.04091854623113389 )*( Seats )  
+ ( -0.09553360817372422 )*( Age_Of_Car )  
+ ( 0.026060011968879804 )*( New_Price_log )  
+ ( -0.025530875779606402 )*( Kilometers_Driven_log )  
+ ( 0.0641071933485399 )*( Location_Bangalore )  
+ ( 0.011103849608016664 )*( Location_Chennai )  
+ ( 0.03381101500736915 )*( Location_Coimbatore )  
+ ( -0.050207434511672436 )*( Location_Delhi )  
+ ( 0.04457203122566265 )*( Location_Hyderabad )  
+ ( -0.047792866099684174 )*( Location_Jaipur )  
+ ( -0.06029914532680927 )*( Location_Kochi )  
+ ( -0.150174336066836 )*( Location_Kolkata )  
+ ( -0.04693452105273331 )*( Location_Mumbai )  
+ ( -0.04327649979405802 )*( Location_Pune )  
+ ( 0.08616622370325164 )*( Fuel_Type_Diesel )  
+ ( -5.221517662690189e-16 )*( Fuel_Type_Electric )  
+ ( -6.661338147750939e-16 )*( Fuel_Type_LPG )  
+ ( -0.08616622370325062 )*( Fuel_Type_Petrol )  
+ ( -0.09432156411944527 )*( Transmission_Manual )  
+ ( 2.3592239273284576e-16 )*( Owner_Type_Fourth & Above )  
+ ( -0.04528670665858933 )*( Owner_Type_Second )  
+ ( 1.1102230246251565e-16 )*( Owner_Type_Third )  
+ ( 0.3031777538724239 )*( Make_Audi )  
+ ( -1.3877787807814457e-15 )*( Make_Bentley )  
+ ( 0.3011491076838147 )*( Make_Bmw )  
+ ( -6.661338147750939e-16 )*( Make_Chevrolet )  
+ ( -0.3755423327481152 )*( Make_Datsun )  
+ ( -0.2936543168648398 )*( Make_Fiat )  
+ ( -6.38378239159465e-16 )*( Make_Force )  
+ ( -0.1682854204695558 )*( Make_Ford )  
+ ( 2.220446049250313e-16 )*( Make_Hindustan )  
+ ( -0.10704863831941518 )*( Make_Honda )  
+ ( -0.12806040598712873 )*( Make_Hyundai )  
+ ( -0.33327382978224224 )*( Make_Isuzu )  
+ ( 0.298837974467263 )*( Make_Jaguar )  
+ ( 0.1563055842979586 )*( Make_Jeep )  
+ ( -1.6653345369377348e-16 )*( Make_Lamborghini )  
+ ( 0.5013595084932326 )*( Make_Land )  
+ ( -0.16479031269514732 )*( Make_Mahindra )  
+ ( -0.10759411549332468 )*( Make_Maruti )  
+ ( 0.3143875839001142 )*( Make_Mercedes-Benz )  
+ ( 0.5404905248544429 )*( Make_Mini )  
+ ( -0.010720315120660935 )*( Make_Mitsubishi )  
+ ( -0.1104368003362399 )*( Make_Nissan )  
+ ( 0.0 )*( Make_Opelcorsa )  
+ ( 0.0 )*( Make_Porsche )  
+ ( -0.1762579116039613 )*( Make_Renault )  
+ ( -0.06270031845288808 )*( Make_Skoda )  
+ ( 0.0 )*( Make_Smart )  
+ ( -0.48580075319889493 )*( Make_Tata )  
+ ( 0.07313438157354718 )*( Make_Toyota )  
+ ( -0.14535156484686274 )*( Make_Volkswagen )  
+ ( 0.18067461677643792 )*( Make_Volvo )  
+ ( 0.3422128551096242 )*( Price_Type_MID-SCALE )  
+ ( 0.6888816750718074 )*( Price_Type_LUXURY )

	count	mean	std	min	25%	50%	75%	max
Year	7253.000	2013.365	3.254	1996.000	2011.000	2014.000	2016.000	2019.000
Kilometers_Driven	7253.000	58699.063	84427.721	171.000	34000.000	53416.000	73000.000	6500000.000
Seats	7200.000	5.280	0.812	0.000	5.000	5.000	5.000	10.000
Price	6019.000	9.479	11.188	0.440	3.500	5.640	9.950	160.000

	count	mean	std	min	25%	50%	75%	max
Year	7253.000	2013.365	3.254	1996.000	2011.000	2014.000	2016.000	2019.000
Kilometers_Driven	7253.000	58699.063	84427.721	171.000	34000.000	53416.000	73000.000	6500000.000
Mileage	7251.000	18.142	4.562	0.000	15.170	18.160	21.100	33.540
Engine	7207.000	1616.573	595.285	72.000	1198.000	1493.000	1968.000	5998.000
Power	7078.000	112.765	53.494	34.200	75.000	94.000	138.100	616.000
Seats	7200.000	5.280	0.812	0.000	5.000	5.000	5.000	10.000
New_Price	1006.000	22.780	27.759	3.910	7.885	11.570	26.043	375.000
Price	6019.000	9.479	11.188	0.440	3.500	5.640	9.950	160.000
Age_Of_Car	7253.000	7.635	3.254	2.000	5.000	7.000	10.000	25.000

	count	mean	std	min	25%	50%	75%	max
Mileage	7019.000	18.210	4.512	0.000	15.290	18.250	21.100	33.540
Engine	7019.000	1613.155	581.855	624.000	1198.000	1493.000	1968.000	5461.000
Power	7019.000	112.134	51.449	34.200	75.000	94.000	138.030	503.000
Seats	7019.000	5.279	0.796	0.000	5.000	5.000	5.000	10.000
Age_Of_Car	7019.000	7.532	3.166	2.000	5.000	7.000	9.000	23.000
Price_log	7019.000	1.840	0.849	-0.821	1.267	1.749	2.303	5.075
New_Price_log	7019.000	2.648	0.828	1.364	2.111	2.411	2.959	5.438
Kilometers_Driven_log	7019.000	10.754	0.714	5.142	10.421	10.878	11.193	13.561

Model	Run Type	Total Features	Size	score %	R-squared	Adj. R-squared	RMSE	MAE	MSE	SST	SSE
With all Features	Training	58	4913	94.138	0.941	0.941	0.204	0.157	0.042	3500.187	205.188
With all Features	Test	58	2106	93.711	0.937	0.935	0.216	0.159	0.047	1558.031	97.983
With all Features with higly sale category	Training	58	4562	94.396	0.944	0.943	0.198	0.151	0.039	3195.380	179.084
With all Features with higly sale category	Test	58	1956	93.206	0.932	0.930	0.211	0.157	0.044	1279.445	86.921

Model	Run Type	Total Features	Size	score %	R-squared	Adj. R-squared	RMSE	MAE	MSE	SST	SSE
With all Features	Training	58	4913	94.138	0.941	0.941	0.204	0.157	0.042	3500.187	205.188
With all Features	Test	58	2106	93.711	0.937	0.935	0.216	0.159	0.047	1558.031	97.983
With all Features with higly sale category	Training	58	4562	94.396	0.944	0.943	0.198	0.151	0.039	3195.380	179.084
With all Features with higly sale category	Test	58	1956	93.206	0.932	0.930	0.211	0.157	0.044	1279.445	86.921
With Model 2 Featuresm, No Power, No New Price	Training	56	4562	93.668	0.937	0.936	0.211	0.162	0.044	3195.380	202.320
With Model 2 Featuresm, No Power, No New Price	Test	56	1956	92.829	0.928	0.926	0.217	0.166	0.047	1279.445	91.754

	Name	Location	Year	Kilometers_Driven	Fuel_Type	Transmission	Owner_Type	Mileage	Engine	Power	Seats	New_Price	Price
S.No.
2397	Ford EcoSport 1.5 Petrol Trend	Kolkata	2016	21460	Petrol	Manual	First	17.0 kmpl	1497 CC	121.36 bhp	5.000	9.47 Lakh	6.000
3777	Maruti Wagon R VXI 1.2	Kochi	2015	49818	Petrol	Manual	First	21.5 kmpl	1197 CC	81.80 bhp	5.000	5.44 Lakh	4.110
4425	Ford Endeavour 4x2 XLT	Hyderabad	2007	130000	Diesel	Manual	First	13.1 kmpl	2499 CC	141 bhp	7.000	NaN	6.000
3661	Mercedes-Benz E-Class E250 CDI Avantgrade	Coimbatore	2016	39753	Diesel	Automatic	First	13.0 kmpl	2143 CC	201.1 bhp	5.000	NaN	35.280
4514	Hyundai Xcent 1.2 Kappa AT SX Option	Kochi	2016	45560	Petrol	Automatic	First	16.9 kmpl	1197 CC	82 bhp	5.000	NaN	6.340
599	Toyota Innova Crysta 2.8 ZX AT	Coimbatore	2019	40674	Diesel	Automatic	First	11.36 kmpl	2755 CC	171.5 bhp	7.000	28.05 Lakh	24.820
186	Mercedes-Benz E-Class E250 CDI Avantgrade	Bangalore	2014	37382	Diesel	Automatic	First	13.0 kmpl	2143 CC	201.1 bhp	5.000	NaN	32.000
305	Audi A6 2011-2015 2.0 TDI Premium Plus	Kochi	2014	61726	Diesel	Automatic	First	17.68 kmpl	1968 CC	174.33 bhp	5.000	NaN	20.770
4582	Hyundai i20 1.2 Magna	Kolkata	2011	36000	Petrol	Manual	First	18.5 kmpl	1197 CC	80 bhp	5.000	NaN	2.500
5434	Honda WR-V Edge Edition i-VTEC S	Kochi	2019	13913	Petrol	Manual	First	17.5 kmpl	1199 CC	88.7 bhp	5.000	9.36 Lakh	8.200

	Name	Location	Year	Kilometers_Driven	Fuel_Type	Transmission	Owner_Type	Mileage	Engine	Power	Seats	New_Price	Price
S.No.
2245	Maruti Wagon R VXI 1.2	Mumbai	2015	55000	Petrol	Manual	First	21.500	1197.000	81.800	5.000	5.880	3.100
6407	Maruti Swift LXI Option	Kochi	2017	62889	Petrol	Manual	First	20.400	1197.000	81.800	5.000	NaN	NaN
1226	Toyota Corolla Altis 1.8 VL CVT	Delhi	2017	21000	Petrol	Automatic	First	14.280	1798.000	138.030	5.000	23.250	18.000
533	Honda City 1.5 V AT	Mumbai	2010	69000	Petrol	Automatic	Second	16.800	1497.000	118.000	5.000	NaN	2.850
6330	Hyundai i20 Asta 1.2	Kolkata	2015	33000	Petrol	Manual	First	18.600	1197.000	81.830	5.000	NaN	NaN
5931	Maruti Ertiga VXI	Mumbai	2012	52500	Petrol	Manual	Second	16.020	1373.000	93.700	7.000	NaN	5.250
2784	Nissan Micra Diesel XV	Hyderabad	2011	86732	Diesel	Manual	First	23.080	1461.000	63.100	5.000	NaN	3.600
5180	Skoda Superb Elegance 1.8 TSI AT	Bangalore	2009	75000	Petrol	Automatic	Second	13.700	1798.000	157.750	5.000	NaN	4.950
2722	Volkswagen Ameo 1.5 TDI Highline	Kochi	2018	31426	Diesel	Manual	First	21.660	1498.000	108.620	5.000	9.720	7.900
5322	Hyundai Grand i10 CRDi Sportz	Ahmedabad	2016	30000	Diesel	Manual	First	24.000	1120.000	70.000	5.000	NaN	5.500

	Name	Location	Year	Kilometers_Driven	Fuel_Type	Transmission	Owner_Type	Mileage	Engine	Power	Seats	New_Price	Price	Make	Model	Age_Of_Car
S.No.
4011	Fiat Punto 1.3 Emotion	Pune	2011	45271	Diesel	Manual	First	20.300	NaN	NaN	NaN	NaN	2.600	Fiat	Punto 1.3 Emotion	10
2074	Maruti Swift 1.3 LXI	Pune	2011	24255	Petrol	Manual	First	16.100	NaN	NaN	NaN	NaN	3.150	Maruti	Swift 1.3 Lxi	10
5185	Maruti Swift 1.3 LXI	Delhi	2012	52000	Petrol	Manual	First	16.100	NaN	NaN	NaN	NaN	3.650	Maruti	Swift 1.3 Lxi	9
2530	BMW 5 Series 520d Sedan	Kochi	2014	64158	Diesel	Automatic	First	18.480	NaN	NaN	NaN	NaN	17.890	Bmw	5 Series 520D Sedan	7
1460	Land Rover Range Rover Sport 2005 2012 Sport	Coimbatore	2008	69078	Petrol	Manual	First	0.000	NaN	NaN	NaN	NaN	40.880	Land	Rover Range Rover Sport 2005 2012 Sport	13
4697	Fiat Punto 1.2 Dynamic	Kochi	2017	17941	Petrol	Manual	First	15.700	NaN	NaN	NaN	NaN	3.930	Fiat	Punto 1.2 Dynamic	4
2668	Maruti Swift 1.3 VXi	Kolkata	2014	32986	Petrol	Manual	First	16.100	NaN	NaN	NaN	NaN	4.240	Maruti	Swift 1.3 Vxi	7
4229	Hyundai Santro Xing XG	Bangalore	2005	79000	Petrol	Manual	Second	17.000	NaN	NaN	NaN	NaN	1.650	Hyundai	Santro Xing Xg	16
3810	Honda CR-V AT With Sun Roof	Kolkata	2013	27000	Petrol	Automatic	First	14.000	NaN	NaN	NaN	NaN	11.990	Honda	Cr-V At With Sun Roof	8
4712	Hyundai Santro Xing XG	Pune	2003	80000	Petrol	Manual	Second	17.000	NaN	NaN	NaN	NaN	0.900	Hyundai	Santro Xing Xg	18

	Name	Location	Year	Fuel_Type	Transmission	Owner_Type	Mileage	Engine	Power	Seats	Make	Model	Age_Of_Car	Price_log	New_Price_log	Kilometers_Driven_log	Price_Type
S.No.
2213	Maruti Swift VXI	Mumbai	2012	Petrol	Manual	First	18.600	1197.000	85.800	5.000	Maruti	Swift Vxi	9	1.295	1.556	9.616	ECONOMY
681	Volkswagen Polo 1.2 MPI Highline	Pune	2017	Petrol	Manual	Second	16.200	1199.000	74.000	5.000	Volkswagen	Polo 1.2 Mpi Highline	4	1.792	2.421	10.897	MID-SCALE
6811	Honda City i VTEC CVT VX	Hyderabad	2015	Petrol	Automatic	First	18.000	1497.000	117.300	5.000	Honda	City I Vtec Cvt Vx	6	2.110	2.416	10.592	MID-SCALE
401	Mitsubishi Pajero Sport 4X2 AT DualTone BlackTop	Hyderabad	2018	Diesel	Automatic	First	13.500	2477.000	175.560	7.000	Mitsubishi	Pajero Sport 4X2 At Dualtone Blacktop	3	3.332	3.524	9.393	LUXURY
6207	Hyundai Verna CRDi ABS	Chennai	2007	Diesel	Manual	First	18.100	1493.000	110.000	5.000	Hyundai	Verna Crdi Abs	14	0.668	2.644	11.238	ECONOMY
5571	Hyundai i20 Asta 1.2	Bangalore	2014	Petrol	Manual	Second	18.600	1197.000	81.830	5.000	Hyundai	I20 Asta 1.2	7	1.732	2.440	10.343	MID-SCALE
6345	Honda Amaze S i-Vtech	Kolkata	2013	Petrol	Manual	First	18.000	1198.000	86.700	5.000	Honda	Amaze S I-Vtech	8	1.335	2.073	11.159	ECONOMY
4055	Mercedes-Benz E-Class E350 CDI	Delhi	2014	Diesel	Automatic	First	13.000	2987.000	261.500	5.000	Mercedes-Benz	E-Class E350 Cdi	7	3.314	4.366	11.127	LUXURY
2799	Mahindra Scorpio VLX 2.2 mHawk BSIII	Delhi	2010	Diesel	Manual	First	12.050	2179.000	120.000	8.000	Mahindra	Scorpio Vlx 2.2 Mhawk Bsiii	11	1.482	2.411	11.184	ECONOMY
529	Toyota Fortuner 3.0 Diesel	Hyderabad	2010	Diesel	Manual	First	11.500	2982.000	171.000	7.000	Toyota	Fortuner 3.0 Diesel	11	2.565	3.366	11.513	MID-SCALE
2510	BMW 3 Series 320d Luxury Line	Mumbai	2014	Diesel	Automatic	Second	22.690	1995.000	190.000	5.000	Bmw	3 Series 320D Luxury Line	7	3.045	3.997	9.831	LUXURY
67	Mercedes-Benz C-Class Progressive C 220d	Coimbatore	2019	Diesel	Automatic	First	0.000	1950.000	194.000	5.000	Mercedes-Benz	C-Class Progressive C 220D	2	3.574	3.895	9.640	LUXURY
2968	Honda Brio S Option MT	Pune	2012	Petrol	Manual	First	18.400	1198.000	86.800	5.000	Honda	Brio S Option Mt	9	0.833	1.890	10.800	ECONOMY
1341	Maruti Ciaz ZDi Plus SHVS	Jaipur	2015	Diesel	Manual	First	28.090	1248.000	88.500	5.000	Maruti	Ciaz Zdi Plus Shvs	6	1.960	1.770	10.609	MID-SCALE
4048	Honda City 1.5 V MT	Delhi	2010	Petrol	Manual	First	17.000	1497.000	118.000	5.000	Honda	City 1.5 V Mt	11	1.253	2.320	11.294	ECONOMY
1939	Mahindra Scorpio S6 7 Seater	Kolkata	2015	Diesel	Manual	Second	15.400	2179.000	120.000	7.000	Mahindra	Scorpio S6 7 Seater	6	2.140	2.457	10.866	MID-SCALE
5639	Volkswagen Polo Petrol Comfortline 1.2L	Kolkata	2013	Petrol	Manual	First	16.470	1198.000	73.900	5.000	Volkswagen	Polo Petrol Comfortline 1.2L	8	1.082	2.317	10.086	ECONOMY
2061	Jeep Compass 2.0 Longitude	Coimbatore	2018	Diesel	Manual	First	17.100	1956.000	170.000	5.000	Jeep	Compass 2.0 Longitude	3	2.947	3.112	10.411	MID-SCALE
658	Toyota Corolla Executive (HE)	Chennai	2007	Petrol	Manual	First	13.400	1794.000	125.000	5.000	Toyota	Corolla Executive (He)	14	1.099	3.073	11.396	ECONOMY
6391	Hyundai i10 Sportz 1.2	Pune	2008	Petrol	Manual	First	20.360	1197.000	78.900	5.000	Hyundai	I10 Sportz 1.2	13	0.693	2.644	11.345	ECONOMY
4319	Honda Brio V MT	Mumbai	2012	Petrol	Manual	First	19.400	1198.000	86.800	5.000	Honda	Brio V Mt	9	1.082	1.890	10.897	ECONOMY
6196	Honda Brio VX	Bangalore	2015	Petrol	Manual	First	19.400	1198.000	86.800	5.000	Honda	Brio Vx	6	1.723	2.416	11.045	MID-SCALE
3844	Hyundai i20 1.2 Sportz	Kolkata	2016	Petrol	Manual	First	17.000	1197.000	80.000	5.000	Hyundai	I20 1.2 Sportz	5	1.658	2.655	9.853	MID-SCALE
4944	Maruti Swift Dzire VDI	Pune	2014	Diesel	Manual	First	23.400	1248.000	74.000	5.000	Maruti	Swift Dzire Vdi	7	1.609	1.909	12.191	MID-SCALE
1989	Hyundai i20 1.4 CRDi Magna	Hyderabad	2012	Diesel	Manual	First	21.900	1396.000	90.000	5.000	Hyundai	I20 1.4 Crdi Magna	9	1.374	2.617	11.175	ECONOMY
5739	Maruti Swift Dzire VDi	Hyderabad	2014	Diesel	Manual	First	19.300	1248.000	73.900	5.000	Maruti	Swift Dzire Vdi	7	1.887	1.909	11.240	MID-SCALE
253	Honda Jazz 1.2 VX i VTEC	Coimbatore	2017	Petrol	Manual	First	18.700	1199.000	88.700	5.000	Honda	Jazz 1.2 Vx I Vtec	4	1.970	2.685	10.811	MID-SCALE
1249	Maruti Swift VXi BSIV	Chennai	2014	Petrol	Manual	Second	16.100	1197.000	85.000	5.000	Maruti	Swift Vxi Bsiv	7	1.504	1.909	10.897	MID-SCALE
425	Hyundai Creta 1.6 SX Plus Dual Tone Petrol	Coimbatore	2017	Petrol	Manual	First	15.290	1591.000	121.300	5.000	Hyundai	Creta 1.6 Sx Plus Dual Tone Petrol	4	2.487	2.083	10.050	MID-SCALE
670	Maruti Ciaz ZXi	Pune	2014	Petrol	Manual	First	20.730	1373.000	91.100	5.000	Maruti	Ciaz Zxi	7	1.792	1.909	10.810	MID-SCALE

	Location	Fuel_Type	Transmission	Owner_Type	Mileage	Engine	Power	Seats	Make	Model	Age_Of_Car	Price_log	New_Price_log	Kilometers_Driven_log	Price_Type
S.No.
4921	Mumbai	Petrol	Manual	First	18.600	1197.000	81.830	5.000	Hyundai	I20 Sportz 1.2	5	1.749	2.655	10.086	MID-SCALE
4525	Pune	Diesel	Manual	First	15.100	2179.000	140.000	7.000	Mahindra	Xuv500 W8 2Wd	9	1.975	2.411	11.462	MID-SCALE
4871	Bangalore	Diesel	Manual	Third	13.100	2499.000	141.000	7.000	Ford	Endeavour 2.5L 4X2	10	1.939	2.241	11.170	MID-SCALE
2527	Pune	CNG	Manual	Third	17.000	1086.000	83.000	5.000	Hyundai	Santro Xing Xo Cng	16	0.182	2.498	11.513	ECONOMY
6071	Mumbai	Diesel	Automatic	First	11.400	2953.000	153.860	7.000	Ford	Endeavour 3.0L 4X4 At	11	1.675	2.241	11.562	MID-SCALE
6508	Bangalore	Diesel	Manual	Second	27.620	793.000	47.000	5.000	Maruti	Celerio Zdi	6	1.281	1.770	9.741	ECONOMY
2117	Kochi	Petrol	Automatic	First	21.400	1197.000	83.100	5.000	Maruti	Baleno Delta Automatic	3	2.066	2.323	9.984	MID-SCALE
135	Kochi	Petrol	Automatic	First	17.000	1591.000	121.300	5.000	Hyundai	Verna Vtvt 1.6 At Sx Plus	2	2.442	2.602	9.445	MID-SCALE
4432	Kolkata	Petrol	Manual	First	22.740	796.000	47.300	5.000	Maruti	Alto 800 Vxi	5	0.892	2.149	10.098	ECONOMY
420	Kochi	Petrol	Manual	First	18.900	1197.000	81.860	5.000	Hyundai	Grand I10 Sportz Edition	3	1.705	2.131	10.605	MID-SCALE

	Mileage	Engine	Power	Seats	Age_Of_Car	New_Price_log	Kilometers_Driven_log	Location_Bangalore	Location_Chennai	Location_Coimbatore	Location_Delhi	Location_Hyderabad	Location_Jaipur	Location_Kochi	Location_Kolkata	Location_Mumbai	Location_Pune	Fuel_Type_Diesel	Fuel_Type_Electric	Fuel_Type_LPG	Fuel_Type_Petrol	Transmission_Manual	Owner_Type_Fourth & Above	Owner_Type_Second	Owner_Type_Third	Make_Audi	Make_Bentley	Make_Bmw	Make_Chevrolet	Make_Datsun	Make_Fiat	Make_Force	Make_Ford	Make_Hindustan	Make_Honda	Make_Hyundai	Make_Isuzu	Make_Jaguar	Make_Jeep	Make_Lamborghini	Make_Land	Make_Mahindra	Make_Maruti	Make_Mercedes-Benz	Make_Mini	Make_Mitsubishi	Make_Nissan	Make_Opelcorsa	Make_Porsche	Make_Renault	Make_Skoda	Make_Smart	Make_Tata	Make_Toyota	Make_Volkswagen	Make_Volvo	Price_Type_MID-SCALE	Price_Type_LUXURY
S.No.
0	26.600	998.000	58.160	5.000	11	1.707	11.184	0	0	0	0	0	0	0	0	1	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
1	19.670	1582.000	126.200	5.000	6	2.180	10.621	0	0	0	0	0	0	0	0	0	1	1	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0
2	18.200	1199.000	88.700	5.000	10	2.153	10.736	0	1	0	0	0	0	0	0	0	0	0	0	0	1	1	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0
3	20.770	1248.000	88.760	7.000	9	1.556	11.374	0	1	0	0	0	0	0	0	0	0	1	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0
4	15.200	1968.000	140.800	5.000	8	4.852	10.613	0	0	1	0	0	0	0	0	0	0	1	0	0	0	0	0	1	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0

	Mileage	Engine	Power	Seats	Age_Of_Car	Price_log	New_Price_log	Kilometers_Driven_log
Mileage	1.000	-0.593	-0.503	-0.321	-0.305	-0.261	-0.422	-0.134
Engine	-0.593	1.000	0.852	0.433	0.048	0.678	0.657	0.160
Power	-0.503	0.852	1.000	0.123	-0.025	0.760	0.765	0.022
Seats	-0.321	0.433	0.123	1.000	-0.014	0.181	0.016	0.185
Age_Of_Car	-0.305	0.048	-0.025	-0.014	1.000	-0.505	-0.008	0.507
Price_log	-0.261	0.678	0.760	0.181	-0.505	1.000	0.706	-0.201
New_Price_log	-0.422	0.657	0.765	0.016	-0.008	0.706	1.000	0.016
Kilometers_Driven_log	-0.134	0.160	0.022	0.185	0.507	-0.201	0.016	1.000

	Model	Run Type	Total Features	Size	score %	R-squared	Adj. R-squared	RMSE	MAE	MSE	SST	SSE
0	With Highly Correlated Features	Training	55	4913	93.980	0.940	0.939	0.207	0.159	0.043	3500.187	210.698
0	With Highly Correlated Features	Test	55	2106	93.550	0.936	0.934	0.218	0.161	0.048	1558.031	100.489

	Coefficients
Intercept	1.968
Mileage	-0.009
Engine	0.000
Power	0.004
Seats	0.041
Age_Of_Car	-0.096
New_Price_log	0.026
Kilometers_Driven_log	-0.026
Location_Bangalore	0.064
Location_Chennai	0.011
Location_Coimbatore	0.034
Location_Delhi	-0.050
Location_Hyderabad	0.045
Location_Jaipur	-0.048
Location_Kochi	-0.060
Location_Kolkata	-0.150
Location_Mumbai	-0.047
Location_Pune	-0.043
Fuel_Type_Diesel	0.086
Fuel_Type_Electric	-0.000
Fuel_Type_LPG	-0.000
Fuel_Type_Petrol	-0.086
Transmission_Manual	-0.094
Owner_Type_Fourth & Above	0.000
Owner_Type_Second	-0.045
Owner_Type_Third	0.000
Make_Audi	0.303
Make_Bentley	-0.000
Make_Bmw	0.301
Make_Chevrolet	-0.000
Make_Datsun	-0.376
Make_Fiat	-0.294
Make_Force	-0.000
Make_Ford	-0.168
Make_Hindustan	0.000
Make_Honda	-0.107
Make_Hyundai	-0.128
Make_Isuzu	-0.333
Make_Jaguar	0.299
Make_Jeep	0.156
Make_Lamborghini	-0.000
Make_Land	0.501
Make_Mahindra	-0.165
Make_Maruti	-0.108
Make_Mercedes-Benz	0.314
Make_Mini	0.540
Make_Mitsubishi	-0.011
Make_Nissan	-0.110
Make_Opelcorsa	0.000
Make_Porsche	0.000
Make_Renault	-0.176
Make_Skoda	-0.063
Make_Smart	0.000
Make_Tata	-0.486
Make_Toyota	0.073
Make_Volkswagen	-0.145
Make_Volvo	0.181
Price_Type_MID-SCALE	0.342
Price_Type_LUXURY	0.689

Cars4U - Price predictions on used cars¶

Objective¶

Key Questions¶

Data Description¶

Data Dictionary¶

Flow & Approach¶

Import necessary python libraries¶

Loading and exploring the data¶

Read Data from csv file and load data into a panda dataframe¶

Common Methods on DataFrame¶

Checking Data¶

Data Shape¶

observations¶

Data Types¶

observations¶

using describe check numerical column stats¶

observation on describe¶

Checking Null Counts¶

observations¶

Check for Duplicates¶

observations¶

Top column values & observations¶

observations¶

observations¶

Data preprocessing with Exploratory Data Analysis & Insights¶

Processing columns - Column Conversions¶

Make columns category type¶

check data types for possible column type conversions¶

possible category columns, save memory usage¶

observations¶

Processing columns - Data Coversions¶

Convert to numerical type columns¶

observations¶

Convert Price and New Price columns¶

observations¶

Feature Engineering - Make new Features¶

Extract Make & Model from Name¶

observations¶

Determine Age of Car using Year¶

observations¶

Initial Univariate analysis¶

Check how Price is distributed¶

Observation on Price¶

Check how New_Price is distributed¶

Observation on New_Price¶

Check how Kilometers_Driven is distributed¶

Observation on Kilometers_Driven - And Fix Extreme Outlier¶

Check how Kilometers_Driven is distributed - after outlier treatment¶

Observation on Kilometers_Driven¶

Missing value Treatment¶

Validate Data¶

Column & Row level missing data¶

check on column level¶

check on row level¶

observations¶

check by missing count¶

Check Existing Data has values for missing fields¶

Fix missing values - Missing value Treatment¶

Fix Engine Values¶

observation¶

Fix Power Values¶

observation¶

Fix Seats Values¶

observation¶

checking missing value counts by columns - after initial treatment¶

observation.¶

Fix Missing price values¶

Observation¶

Fix new_price¶

observation¶

Outlier Treatment - Fix Outlier(s)¶

Outlier detection using IQR¶

observations on outliers¶

Data Distribution¶

Log transformation to fix skewness¶

Observation¶

Check once for any missing values after all treatment(s)¶

Observations missing data¶

Drop missing values¶

No missing values¶